Evaluating supervised concept drift detection for online imputation on wearable data streams with changing missing mechanisms

• Afonso M. S. Lima Universidade de São Paulo (USP)
• Matheus Q. Mota Universidade de São Paulo (USP)
• Leonardo S. Cardoso Universidade de São Paulo (USP)
• Elaine P. M. de Sousa Universidade de São Paulo (USP)

DOI: https://doi.org/10.5753/sbbd.2026.249498

Resumo

Wearable data streams often contain missing values, for which there is uncertainty about the cause of the missingness. Further, this missingness may also change throughout the stream processing. Our study evaluates the effectiveness of concept drift detectors applied in a supervised manner for online imputation on heart rate data streams collected from wearable devices. Experiments were conducted under changing missing mechanisms scenarios, utilizing the Hoeffding Adaptive Tree regressor alongside three concept drift detectors for monitoring model error: Page-Hinkley, Kolmogorov-Smirnov Windowing, and Adaptive Windowing. We also compared it with baselines Hoeffding Tree and mean imputation. Our results show that supervised concept drift detection yields worse performance than non-aware concept drift approaches. This indicates that supervised detectors may be insufficient to address data distribution changes caused by missing values, highlighting the need for further research.

Palavras-chave: Data Streams, Imputation, Concept Drift, Online Machine Learning

Referências

Bartz, E. and Bartz-Beielstein, T. (2024). Online Machine Learning. Springer.

Benhamza, K., Benselim, R., Naidja, H., and Seridi, H. (2025). A comprehensive survey of imputation methods in medical missing data analysis. Applied Intelligence, 55(11):778.

Bifet, A., Gavalda, R., Holmes, G., and Pfahringer, B. (2023). Machine learning for data streams: with practical examples in MOA. MIT press.

Isgut, M., Gloster, L., Choi, K., Venugopalan, J., and Wang, M. D. (2022). Systematic review of advanced ai methods for improving healthcare data quality in post covid-19 era. IEEE Reviews in Biomedical Engineering, 16:53–69.

Ismail Fawaz, H., Forestier, G., Weber, J., Idoumghar, L., and Muller, P.-A. (2019). Deep learning for time series classification: a review. Data Mining and Knowledge Discovery, 33(4):917–963.

Lima, A. M. S. (2026). The impact of missing values on anomaly detection tasks in healthcare wearable data. Journal of Information and Data Management. Accepted for publication; to appear.

Lima, A. S. (2025). Assessing the impact of missing value mechanisms on anomaly detection in healthcare wearable data. In Anais do XL Simpósio Brasileiro de Bancos de Dados, pages 781–787, Brasil.

Lima, A. S. and Sousa, E. (2024). Handling missing values in data streams: An overview. In Anais do XXXIX Simpósio Brasileiro de Bancos de Dados, pages 750–756, Brasil.

Lukats, D., Zielinski, O., Hahn, A., and Stahl, F. (2025). A benchmark and survey of fully unsupervised concept drift detectors on real-world data streams. International Journal of Data Science and Analytics, 19(1):1–31.

Mahdi, O. A., Ali, N., Pardede, E., Alazab, A., Al-Quraishi, T., and Das, B. (2024). Roadmap of concept drift adaptation in data stream mining, years later. IEEE Access, 12.

Mangussi, A. D., Santos, M. S., Lopes, F. L., Pereira, R. C., Lorena, A. C., and Abreu, P. H. (2024). mdatagen: A python library for generating missing data. [link].

Mishra, T., Wang, M., Metwally, A. A., Bogu, G. K., Brooks, A. W., Bahmani, A., Alavi, A., Celli, A., Higgs, E., Dagan-Rosenfeld, O., et al. (2020). Pre-symptomatic detection of covid-19 from smartwatch data. Nature biomedical engineering, 4(12):1208–1220.

Montiel, J., Halford, M., Mastelini, S. M., Bolmier, G., Sourty, R., Vaysse, R., Zouitine, A., Gomes, H. M., Read, J., Abdessalem, T., et al. (2021). River: machine learning for streaming data in python. Journal of Machine Learning Research, 22(110):1–8.

Ren, L., Wang, T., Seklouli, A. S., Zhang, H., and Bouras, A. (2023). A review on missing values for main challenges and methods. Information Systems, page 102268.

Santos, M. S., Pereira, R. C., Costa, A. F., Soares, J. P., Santos, J., and Abreu, P. H. (2019). Generating synthetic missing data: A review by missing mechanism. IEEE Access, 7:11651–11667.